How are AI-generated images useful in real projects?
From concept to reality. How AI helped me (or not) create a Promotional game.
By the end of 2022, AI was in the spotlight, and like many people, I was skeptical and rigorous about how it could change us as humans. But all ofmy hesitation vanished when Chat GPT was launched. In just a few weeks, I saw how this AI chatbot could provide valid and valuable responses, and my perception of AI shifted utterly. Let’s explore how AI-generated images could help us.
Why aren’t we using AI-generated images? Is Midjourney our far-off future, or should we demystify It?
There was a sea of articles about Midjourney and other AI-generating tools every hour on Medium. I couldn’t help but notice how excited and impressed the writers were. But, despite all the positive feedback, I wondered why people were not using AI-generated images in their work. I have seen countless AI-generated avatars, Dribbble shots, logos, and even online courses using these programs.
So, why weren’t people implementing them into their projects? Upon further research, I couldn’t find any other examples of AI-generated images used in February 2023 besides a single example:
From concept to reality: How AI helped (or not) create a promo game
I conducted an experiment with implementing AI-generated images to an actual project. I am a Senior Product designer at a gambling product called Leon. My aim was to increase the probability of second deposits from our customers. After conducting extensive research and validating our hypotheses, we launched a promotional game with a Fortune wheel, a well-known mechanism that tests your luck in the gambling industry.
In short, our adrenaline-driven users could spin the wheel and win an award within the first seven days after registering, but they could only claim the prize by making a deposit. Thus, the customer development team believed it would increase our users’ engagement and activity.
After getting approval from our stakeholders, I requested graphics from our design team. However, due to tight deadlines on other projects, I decided to try something new and work with Midjourney instead. I had never used this tool before but was determined to try it.
Brainstorming machine
As a result-driven person, I was excited when I typed all the inputs into Midjourney’s prompt and received an impressive result. I had given information such as the Fortune wheel, 8–12 sections, gambling theme, and our brand’s lion logo. The result generated by AI aligned perfectly with my vision.
Breaking down the wheel
Drawing from my experience in graphic design, I decided to break down the fortune wheel into three components: the wheel (to be spun and animated), the arrow (indicating the selected gift), and the lion (a simple yet effective centerpiece). The challenge now was to create an image that could be easily be divided into these components. It wasn’t easy. Some images had too large lions, while others didn’t match the style I was aiming for.
Aesthetic challenges and learning resources
I taught the AI to understand the aesthetics of online gambling games, which often feature ancient themes and a lot of gold. I wanted the AI to generate images without any tilt or distortion, and I took several attempts to get it right. I tried combining our branding images, but Midjourney didn’t understand my vision.
I had to experiment with different tools and resources, such as Prompt Noonshot and watched YouTube PANTER channel tutorials, to understand how Midjourney’s properties work.
Finding the right fit
After multiple rounds of generation and mixing ideas, I finally found the right fit and was ready to proceed. However, I encountered another obstacle when adjusting the number of sections on the wheel. The AI kept adding more and more cells, while I wanted fewer. I eventually decided to generate the wheel separately and incorporate it into the image, but this proved to be a complex task.
Furthermore, my product’s focus on gambling made it a bit controversial, and I suspected that AI would struggle to provide casino-related elements such as cards, chips, or roulette. My hypothesis was accurate, and all my attempts to generate such prompts were unsuccessful.
The result?
After over 100+ prompts, I finally settled on my favorite combinations and was ready to create a real game. While I don’t believe the result was excellent, it was a great starting point for my graphic design team to work with. As a result, I couldn’t have accomplished this task without the help of graphic designers.
The potential of AI in design: saving time and personalizing
However, AI struggles with emotional nuances, creating original content, and moral norms. Thus, I had to manually edit some of the images in Photoshop to get what I needed.
Nevertheless, I still believe that AI is the future of design. While a designer needs empathy and logical thinking to create convenient and functional projects, AI can be an additional tool to help overcome the fear of a “white sheet” when starting a new project. The neural network can generate several ideas on request, from which designers can take inspiration, but it won’t do all the work for them.
Overall, I am excited to see how AI will continue to evolve and transform the design industry. It may not be perfect yet, but it has tremendous potential to save designers time, personalize content, and create variations, ultimately helping them to create better designs.
Thank you for taking the time to engage with my thoughts.
